Early post-marketing safety signals of CRISPR-edited exagamglogene autotemcel (Casgevy) versus lentiviral betibeglogene autotemcel (Zynteglo): A faers disproportionality study, 2014–2025

Abstract

Abstract Background: CRISPR-based genome editing entered clinical use with exagamglogene autotemcel (Casgevy) for severe sickle cell disease and transfusion-dependent β-thalassemia, approved by the U.S. FDA in December 2023. Its lentiviral predecessor, betibeglogene autotemcel (Zynteglo), was approved in 2022. Randomized trials for both autologous gene therapies enrolled small cohorts, and long-term pharmacovigilance data remain limited. Early identification of post-marketing adverse event (AE) patterns is therefore critical to guide clinicians, regulators, and patients on emerging safety risks, whether unique to CRISPR editing or shared with intensive myeloablative transplantation. Methods: We queried FDA Adverse Event Reporting System (FAERS) files from Q1-2014 through Q1-2025 using R (v4.4.0). Individual case safety reports (ICSRs) listing Casgevy or Zynteglo as the primary suspect were identified by querying drugname and prod_ai fields for “exagamglogene,” “Casgevy,” “betibeglogene,” or “Zynteglo.” Preferred Terms (PTs) from REAC tables were manually grouped into clinically relevant buckets: Bleeding/Vascular, Gastrointestinal/Mucosal, Haematologic/Organs, Infection/Inflammatory, Liver/Metabolic, Immune/Other, Neurologic, and Other/Miscellaneous. Buckets were included if they contained at least three PTs for either product. We computed crude reporting odds ratios (RORs) and 95% Wald confidence intervals (CIs), applying a 0.5 continuity correction. The comparator was all other FAERS primary suspect drug–event combinations during the study period, excluding any “autotemcel” products. Results: We identified 33 primary suspect ICSRs: 5 for Casgevy (median age 18 years; interquartile range [IQR] 18–18; 0% female) and 28 for Zynteglo (median age 10 years; IQR 7–13; 32% female). No reports were designated as fatal or life-threatening. Casgevy generated 9 PT entries, most frequently in the Gastrointestinal/Mucosal bucket (4/9), yielding an ROR of 29.2 (95% CI, 8.4–101.3). Bleeding/Vascular events (2/9) showed an ROR of 65.0 (15.5–272.0). Single reports in Haematologic/Organs, Infection/Inflammatory, and Neurologic categories produced elevated point estimates with wide intervals (Haematologic/Organs ROR 78.0; Infection/Inflammatory ROR 30.1; Neurologic ROR 17.1). Zynteglo contributed 91 PTs distributed across seven buckets. Bleeding/Vascular events were most common (n = 32; ROR 106.5, 69.5–163.3), followed by Haematologic/Organs (n = 7; ROR 39.2, 18.6–82.8) and Infection/Inflammatory (n = 6; ROR 12.9, 5.8–28.7). Liver/Metabolic (n = 2; ROR 187.7, 53.4–659.6) and Immune/Other (n = 2; ROR 118.8, 33.8–417.6) signals were of high magnitude but limited interpretability due to sparse counts. Conclusions: This first FAERS-based safety analysis of CRISPR-edited therapy suggests Casgevy may be associated with early post-infusion gastrointestinal/mucosal and bleeding events, while Zynteglo shows a broader haematologic and vascular signal profile, with isolated liver and immune AEs. These disproportionate reporting signals, although preliminary, support enhanced clinical vigilance, especially for mucosal toxicity and bleeding. Limitations include small case counts, underreporting biases in FAERS, absent exposure denominators, and potential misclassification from manual PT grouping. These findings generate biologically plausible hypotheses and underscore the need for prospective real-world safety data as uptake of CRISPR and lentiviral gene therapies expands.